American Journal of Health Education

ISSN: 1932-5037 (Print) 2168-3751 (Online) Journal homepage: https://tandfonline.com/loi/ujhe20

Does More Sleep Time Improve ? Evidence for the Middle-Aged and Elderly

Wei Chen

To cite this article: Wei Chen (2019) Does More Sleep Time Improve Memory? Evidence for the Middle-Aged and Elderly, American Journal of Health Education, 50:6, 366-373, DOI: 10.1080/19325037.2019.1662859 To link to this article: https://doi.org/10.1080/19325037.2019.1662859

Published online: 03 Oct 2019.

Submit your article to this journal

Article views: 19

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at https://tandfonline.com/action/journalInformation?journalCode=ujhe20 AMERICAN JOURNAL OF HEALTH EDUCATION 2019, VOL. 50, NO. 6, 366–373 https://doi.org/10.1080/19325037.2019.1662859

Does More Sleep Time Improve Memory? Evidence for the Middle-Aged and Elderly Wei Chen Lewis University

ABSTRACT ARTICLE HISTORY Background: The world population is aging rapidly. However, no study exists that examines Received 3 June 2019 specifically the effect of sleep time on memory among the middle-aged and elderly. Purpose: This Accepted 5 August 2019 paper examines the effect of sleep time on memory for the middle-aged and elderly. Methods: Data from the China Health and Retirement Longitudinal Study (CHARLS) were analyzed using fixed-effects models for panel data. The final sample used for this study included 4,330 middle- aged and elderly adults, of which about 50.4% were male. Results: The effect of sleep time on memory is heterogeneous across gender. One more hour of sleep time per night was estimated to increase the probability of “good memory” by roughly 0.2% (p > .1) for men and 1.6% (p < .01) for women. The odds ratios of “good memory” for men and women were estimated to be around 1.05 (p > .1) and 1.6 (p < .01), respectively. Discussion: This study suggests that efforts aimed at improving sleep time can bring significant memory-related benefit to middle-aged and elderly women. Translation to Health Education Practice: Health care providers and health educators should play a role in raising middle-aged and elderly women’s general awareness of the memory- related benefit of sleep.

Background emotional experiences and self-evaluations.4 Sleep is an important lifestyle factor that affects health. Admittedly, there are many factors that affect an indi- It has been established in the literature that optimizing vidual’s memory. Studies have shown that sleep is sleep duration and quality can be an intervention to important for cognitive performance and memory improve glucose control in patients with type 2 consolidation.5,6 Memory scholars divide memory into diabetes.1 Sleep is shown to affect energy balance, and declarative (explicit) memory and procedural (implicit) more sleep is recommended to prevent obesity.2 While memory. Declarative memory includes episodic mem- affects sleep, sleep restoration is associated ory that is related to the recollection of specific events with decreased severity of depression.3 Given the role and that is associated with the long- of sleep in a person’s daily life and the fact that the term of ideas and concepts.7 Sleep not only aids potential health benefits of sleep have not been fully the consolidation of ,8 but also facil- explored yet, it is no wonder that sleep has also itates the consolidation of semantic memory.9 attracted and received interest from mental refers to the unconscious memory health researchers. of skills. Sleep is found to support the consolidation of As an important part of mental health, memory procedural memory.10 reflects brain functions, influences human behaviors, According to the International Population Reports and impacts quality of life. After all, intelligence is released by the United States Census Bureau (see An essentially a memory-based process. Complaints of Aging World: 2015 for details),11 the world population memory failure are common in clinics and hospital is aging rapidly. Median age is predicted to rise con- settings. Age-related memory impairment (AMI) or tinuously in both developed and developing countries. age-associated memory impairment (AAMI) is China, the country with the largest population in the a natural process related to normal aging. However, it world, is no exception, and it is expected to experience can result in meaningful impacts on a person’s life, a huge increase in the proportion of middle-aged and which include reduced work efficiency and negative elderly people in the future. Provision of health

CONTACT Wei Chen [email protected] Department of Economics, Lewis University, SB-154-N Saint Charles Borromeo Center, One University Parkway, Romeoville, IL 60446 Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/ujhe. © 2019 SHAPE America SLEEP TIME AND MEMORY 367 promotion and wellness services for the middle-aged two years. Four waves had been conducted until 2018 and elderly population is a common challenge facing (2011 wave 1, 2013 wave 2, 2014 wave 3, and 2015 different countries. wave 4). The household response rate exceeded 80% Studies in the literature generally focus on the effects in all waves. Participants gave written informed con- of sleep in adolescents or the whole adult sent before they participated in the survey in each population.12,13 Only a few studies exist that target the wave. middle-age and elderly specifically. Health education A specially designed Computer-Assisted Personal and health promotion targeting older people differ sig- Interviewing (CAPI) system was used to perform qual- nificantly from that targeting younger population.14 It ity checks. More specifically, Global Positioning System is thus important to study the effect of sleep time for (GPS) matching was used to ensure the sampled house- the middle-aged and elderly. Researchers have found holds were accurately located. Excessively missing data that midlife and late-life sleep disturbances are asso- and short interviews were checked by a programmer in ciated with a higher dementia risk,15 and that sleep the CHARLS central headquarters when the survey was deprivation may be linked to risk factor for being conducted by two field interviewers, who also Alzheimer’s disease.16 However, the direct effect of provided their personal observations about the inter- sleep time on memory for the middle-aged and elderly viewing process and the respondent’s attitude and com- is still unclear, which remains to be explored. prehensive ability. Sound recording was checked to ensure survey questions were asked correctly and well. Households were called back by central office staff if Purpose technical issues existed that affected the sound record- The purpose of this study was to examine the effect of ing checking. sleep time on memory for the middle-aged and elderly The CHARLS was originally designed based on the and identify possible gender heterogeneity. To the best of Health and Retirement Study (HRS), the English my knowledge, no study exists that examines specifically Longitudinal Study of Aging (ELSA), and the Survey the effect of sleep time on memory among the middle- of Health, Aging and Retirement in Europe (SHARE). aged and elderly. More importantly, panel inference that A pilot survey was conducted in 2008 in two provinces can provide more robust results than cross-sectional stu- (Gansu and Zhejiang), after which the pilot data were dies would allow is sparse in the related literature. This analyzed to improve the quality of the questionnaire. In paper aims to fill in the gap. Based on the empirical 2010, two additional formal pretests were deployed in evidence from China, this study provides implications two areas (Beijing and Langfang) and further modifica- for health education and health care professionals. tions were made based on the pretests before the ques- Possible reasons for the heterogeneous effect of sleep tionnaire was finalized and utilized in all waves. time on memory are discussed for future research to Information collected in CHARLS ranges from socio- focus on. economic status to health conditions. With its extensive coverage and high accuracy, the CHARLS has been used by researchers to study aging,17,18 the effects of social Methods policies and programs,19,20 and the relationships between lifestyles and health.21,22 Intermsofsleep,whilethere Sampling are studies that examine the impacts of sleep on depres- Data used in this study were from the China Health sion, grip strength, and chronic diseases based on the – and Retirement Longitudinal Study (CHARLS). The CHARLS data,23 25 the effect of sleep time on memory goal of CHARLS was to provide high quality data has not been investigated so far. that would meet the needs of scientific research on Since wave 3 (2014) of the CHARLS was a special the middle-aged and elderly, when the rapid aging of survey of life history, it was not used in this study. The Chinese population called for a national project that final sample used for this study consisted of 4,330 could provide sufficient micro-level longitudinal data middle-aged and elderly adults observed in all three for studies on aging problems. Officially started in waves. About 50.4% were male, and the average age 2011, the CHARLS sampled Chinese residents aged was around 63 years. 45 and above randomly selected from 150 counties/ districts and 450 villages/resident committees based Measures and covariates on the multi-stage stratified probability proportional to size (PPS) sampling. The baseline households and Participants were asked to report their demographic infor- individuals surveyed in 2011 were followed up every mation and their memory status based on a five-point 368 W. CHEN

Likert item: “excellent,”“very good,”“good,”“fair” or number of grandchildren was controlled for to capture “poor” in each wave. I created a binary outcome indicator the effect of grandparenting on memory. Living distance that equaled 1 if memory status was reported to be “excel- from children reflected the health care received from lent,”“very good,” or “good,” and 0 otherwise. The binary children and the empty-nest-related health effect (the outcome indicator enabled the probability of “good mem- effect of children’s departure on parents’ health). ory” to be estimated based on the fixed-effects models. Whether father/mother was still living captured the Since the reported memory status was the overall memory effect of caring for aging parents on emotional distress status, the outcome variable measured both declarative and mental health. Because an individual’smemorysta- memory and procedural memory. Participants also tus is affected by his or her overall health, own overall reported their average sleep time per night during the past health was controlled for in the robustness check step. month. Sleep time was defined as the amount of time they Summary statistics are displayed in Table 1. actually slept, not the total time spent in bed per night. Figure 1 shows the relationship between memory Econometric analysis status and sleep time by waves. Sleep time seems to be positively correlated with memory status. To estimate the effect of sleep time on memory for the The covariates in this study included age, gender, middle-aged and elderly, I first applied the following marital status, number of grandchildren, living distance baseline model to the data: from children, and whether father/mother was still liv- 0 Y ¼ β þ β S þ Xitβ þ αi þ μ þ εit (1) ing. Normal aging is associated with a decline in various it 0 1 it 2 t memory abilities, and its effect was captured by the where Yit is the binary indicator for memory status; Sit inclusion of age. Expecting that there is a nonlinear is the average hours of reported sleep time per night relationship between normal aging and memory, I also during the past month, i.e., the average quality sleep added age squared and age cubed as control variables. time that a participant got per night; Xit contains the Marital status was controlled for to reflect lifestyles. aforementioned control variables; αi is the individual- 26 μ ε Since grandparenting affects cognitive health, the fixed effects; t is the time-fixed effects; and it is the

Figure 1. Memory status vs. sleep time by waves (men and women combined). Notes: “1” indicates “good memory” (“excellent,”“very good,” or “good”) and “0” indicates the opposite (“fair” or “poor”). SLEEP TIME AND MEMORY 369

Table 1. Summary statistics. models with and without individual-specific time Variable Mean SD Min. Max. trends. STATA software (version 15) was used to con- Memory status 0.164 0.370 0 1 duct all the analyses. It is worth noting that statistically Sleep time (in hours) 6.188 1.957 0 15 Age 62.604 9.667 45 101 speaking, although the fixed-effects logit models use Gender (1 = male; 0 = female) 0.504 0.500 0 1 only within-individual differences, they are less vulner- Marital status 0.681 0.466 0 1 Number of grandchildren 3.864 3.641 0 18 able to omitted variable bias. Odds-ratios are more Living distance from children (in 36.413 277.022 0 10199 informative than marginal effects estimated from the kilometers) 1 Father still living (1 = yes; 0 = no) 0.100 0.300 0 1 fixed-effects logit models. Mother still living (1 = yes; 0 = no) 0.190 0.392 0 1 Overall health 0.224 0.417 0 1 Notes: Marital status = 1 if married and/or living with a partner; marital status = 0 if otherwise. Results If the participant had more than one child, the average living distance from all children was used. Table 2 shows the estimated effect of sleep time on Overall health = 1 if overall health was reported to be “excellent,”“very good,” or “good”; overall health = 0 if overall health was reported to be memory for the middle-aged and elderly (men and “fair” or “poor.” women combined). Adding individual-specific time trends did not significantly change the magnitudes of β β^ random error. 1, the coefficient of Sit, is the measure 1, the estimated effect of sleep time on memory. One of the effect of sleep time on memory. more hour of sleep time per night was estimated to Individual-fixed effects controlled for omitted vari- increase the probability of “good memory” by roughly ables that were constant over time but varied across 1% for the middle-aged and elderly. The results from individuals (e.g., individual genetic characteristics), and the fixed-effects logit models are listed in column (3) time-fixed effects controlled for omitted variables that and column (4) of Table 2. The estimated coefficients were constant across individuals but varied over time of sleep time from the baseline and extended fixed- (e.g., improvements of medical technologies and med- effects logit models were positive and significantly ical services). It is possible that some omitted variables different from zero, indicating that sleep time per might vary both across individuals and over time (e.g., night had a significant positive effect on memory. individual memory loss associated with normal aging The estimated odds ratios were significantly greater and cardiovascular risk factors). Following the method than 1 (p < .01), suggesting that one more hour of 27,28 γ commonly used in the literature, I added i t in sleep time per day increased the odds of having the baseline model to capture the long-term linear a “good memory.” More sleep time per night was trends in memory status that might vary across indivi- shown to help the middle-aged and elderly achieve duals. Equation (2) is the extended model with indivi- a “good memory.” dual-specific time trends: An individual’s sleep time per night is affected by his 0 or her overall health (e.g., whether the individual has Y ¼ β þ β S þ X β þ α þ μ þ γ t þ ε (2) it 0 1 it it 2 i t i it mental illness or depression), and one’s overall health Since sleep time might have a non-linear relationship affects an individual’s memory. Individual-fixed effects, with memory, I also estimated the fixed-effects logit time-fixed effects, and individual-specific time trends

Table 2. The effect of sleep time on memory. (5) (6) (1) (4) Fixed-effects Fixed-effects Fixed-effects (2) Fixed-effects (3) Fixed-effects Fixed-effects logit linear model logit model Estimate linear model linear model logit model model (Robustness test) (Robustness test) Coefficient of sleep time 0.011*** 0.010** 0.122*** 0.300*** 0.009*** 0.291*** (0.002) (0.005) (0.027) (0.080) (0.005) (0.081) Coefficient of overall health –– ––0.120*** 1.080*** (0.022) (0.297) Individual fixed effects Y Y Y Y Y Y Time fixed effects Y Y Y Y Y Y Individual-specific time NY NYYY trends Odds ratio ––1.130*** 1.350*** – 1.337*** (0.030) (0.108) (0.108) Notes: Figures in parentheses are cluster-robust standard errors. Y = Yes; N = No. ** Significant at p < .05 level; ***Significant at p < .01 level. 370 W. CHEN are supposed to solve the omitted variable (OV) Table 4. The effect of sleep time on memory (fixed-effects logit). bias. To test the robustness of the results from the Estimate Men Women fixed-effects linear models and logit models, I added Coefficient of sleep time 0.066* 0.044 0.184*** 0.493*** (0.035) (0.036) (0.042) (0.161) individual overall health to the regression models with Coefficient of overall health – 0.915*** – 1.467*** individual-specific time trends. (0.135) (0.503) Overall health NYNY Column (5) and column (6) of Table 2 contain the Individual fixed effects YYYY estimated effect of sleep time on memory after indivi- Time fixed effects YYYY Individual-specific time trends Y Y Y Y dual overall health was controlled for. Individual over- Odds ratio 1.068* 1.045 1.203*** 1.638*** all health positively and significantly affected an (0.037) (0.038) (0.051) (0.263) ’ Notes: Figures in parentheses are cluster-robust standard errors. Y = Yes; N = No. individual s memory status (p < .01). However, adding *Significant at p < .1 level; ***Significant at p < .01 level. overall health, a factor that significantly affects memory status, did not significantly change the estimated effect of sleep time on memory. The robustness check further However, the estimates of interest were close in magni- confirmed the validity of the results in this study. The tudes in all the baseline and extended models regardless OV bias was resolved by adding individual-fixed effects, of whether overall health was included or not, indicating time-fixed effects, and individual-specific time trends to that the results were consistent in this study. Since over- the models, and the robust effect of sleep time on all health was shown to be an important factor signifi- memory was thus identified. cantly affecting memory status, the estimates from the It is possible that the effect of sleep time on memory models that controlled for overall health, individual- is heterogeneous across gender. Even though gender fixed effects, time-fixed effects, and individual-specific was controlled for in the analyses based on the pooled time trends were the most precise. While sleep time only sample, the coefficient of gender could not capture the had a small positive effect on memory for middle-aged heterogenous effect of sleep time on memory because it and elderly men, it had a large and significant positive only controlled for the gender difference in memory effect for middle-aged and elderly women. The signifi- status not associated with sleep time. To investigate if cant estimated effect of sleep time on memory for the heterogeneity exists, I applied the regression models to middle-aged and elderly based on the pooled sample the subsamples that included only men and only (men and women combined) was driven by the signifi- women. The covariate “gender” was then dropped cant effect among middle-aged and elderly women. from the regression models. Results from the fixed- Specifically, one more hour of sleep time per night was “ ” effects linear models and the fixed-effects logit models estimated to increase the probability of good memory are listed in Tables 3 and 4 respectively. by roughly 0.2% (p > .1) for men and 1.6% (p < .01) for “ ” The estimated effect of sleep time on memory for women (Table 3). The odds ratios of good memory for men was slightly significant (only at 10% significance men and women were estimated to be around 1.05 level) when overall health was not controlled for. (p > .1) and 1.64 (p < .01), respectively (Table 4). The estimated odds ratios suggested that if a middle-aged and elderly man had one more hour of sleep time per “ ” Table 3. The effect of sleep time on memory (fixed-effects night, his odds of having a good memory would only linear). change slightly. However, if a middle-aged and elderly A: The effect of sleep time on memory for men woman had one more hour of sleep time per night, his Coefficient of sleep time 0.006* 0.005 0.003 0.002 or her odds of having a “good memory” would be multi- (0.004) (0.004) (0.007) (0.007) Coefficient of overall health – 0.128*** – 0.115*** plied by roughly 1.64. Put differently, female individuals (0.018) (0.032) with more hours of sleep were more likely to have Overall health N Y N Y “ ” Individual fixed effects Y Y Y Y a good memory . Time fixed effects Y Y Y Y Individual-specific time trends N N Y Y B: The effect of sleep time on memory for women Discussion Coefficient of sleep time 0.014*** 0.013*** 0.017*** 0.016*** (0.003) (0.003) (0.006) (0.006) Coefficient of overall health – 0.115*** – 0.121*** I investigated the effect of sleep time on memory for the (0.016) (0.030) middle-aged and elderly in this study. The panel infer- Overall health N Y N Y Individual fixed effects Y Y Y Y ence enabled the omitted confounders that affected Time fixed effects Y Y Y Y memory status to be controlled for, which enhanced Individual-specific time trends N N Y Y the robustness of the results from this study. Both the Notes: Figures in parentheses are cluster-robust standard errors. Y = Yes; N = No. linear models and non-linear models were applied to the *Significant at p < .1 level; ***Significant at p < .01 level. data, and they yielded consistent results. One more hour SLEEP TIME AND MEMORY 371 of sleep time per night was estimated to increase the during the past month. Using clinical data would possi- probability of “good memory” by roughly 1% for the bly strengthen the validity of the results. However, since middle-aged and elderly as a whole, but the effect of this is the first study to provide evidence of the effect of sleep time on memory was found to be heterogeneous sleep time on memory for the middle-aged and elderly across gender. It should be emphasized that the sleep and the estimates from the baseline and extended mod- time measured in this study was the amount of time els (both linear and nonlinear) were consistent, it lays participants actual slept, not the total time spent in bed. foundation for future studies that rely on clinical Although sleep time only has a small positive effect for data. Second, the results from this study do not indicate middle-aged and elderly men, it has a large and signifi- whether sleep time has a larger effect on declarative cant positive effect for middle-aged and elderly women. memory or procedural memory. Even so, this study This conclusion is consistent with findings in the litera- provides important policy implications for health educa- ture that gender differences exist in sleep health,29,30 and tors. After all, an individual’s life is affected by his or her that sleeping boosts women’s brain power more.31 overall memory status, which includes both declarative While there is a clear need for future research to memory and procedural memory. Third, this study does identify the exact reason(s) for the heterogenous effect not identify which stage of the night sleep is the most of sleep time on memory, this paper provides three crucial for memory improvement, which is another topic possible reasons for future studies. for future studies to investigate. First, since women tend to multi-task and their brains are more complex than men’s, women’s sleep 31 need is greater. It has been shown empirically that Translation to Health Education Practice women report longer ideal sleep duration than men.32 The greater need for sleep may be one possible reason The effect of sleep time on memory was found to be for the greater effect of sleep time on memory for heterogeneous across gender for the middle-aged and women. elderly. This study suggests that efforts aimed at improv- Second, studies have found that there are gender ing sleep time among the middle-aged and elderly, such as differences in circadian rhythms and slow-wave activity community noise control during nighttime and provision (SWA), which affect the homeostatic component of of facilities and amenities that help improve sleep hygiene, .33,34 The greater effect for women found in can lead to improvement in memory for middle-aged and this study may also be attributable to the fact that elderly women, although the effect can be insignificant for women have different circadian regulation of sleep middle-aged and elderly men. Social supports to improve and a greater circadian modulation of cognition due sleep hygiene that target middle-aged and elderly women to gender differences in melatonin amplitude. This should be advocated by health care providers and health explanation echoes the finding that women tend to educators. At the same time, health care providers and rebound more quickly and get greater improvements health educators should play a role in raising middle-aged when they get more restorative sleep.35 and elderly women’s general awareness of the memory- Third, compared with men, women exhibit greater related benefit of sleep and discouraging bad sleep habits signal intensity changes, which are related to neuronal that affect sleep time. Encouraging middle-aged and activities and engagement.36,37 The greater signal intensity elderly women to form a habit of tracking sleep time changes possessed by women may be another possible every day can be a good way of promoting self- explanation for the heterogenous effect of sleep time. management that helps improve memory. Finally, health care providers and health educators should be aware that although taking medications helps reduce the symptoms Strengths and limitations of some chronic diseases, if medications negatively affect One notable strength of this study is the use of panel sleep time, sleep time may take a heavier toll on middle- inference based on the fixed-effects models. The fixed- aged and elderly women’s memory. effects models enabled the omitted confounders to be controlled for, which enhanced the robustness of the Note estimated effect of sleep time on memory. Another strength of this study is the large sample size that 1. The odds of “good memory” were defined as provided enough statistical power to justify the results. ProbabilityðÞ good memory This study has several limitations. First, the analyses 1 ProbabilityðÞ good memory were conducted based on self-reported memory status and self-reported average hours of sleep time per night and the corresponding odds ratio was equal to 372 W. CHEN

Odds of good memory ðif quality sleep time 8. Van der Helm E, Gujar N, Nishida M, Walker MP, per night increases by one hourÞ Robertson E. Sleep-dependent facilitation of episodic Odds of good memory ðif quality sleep time memory details. PLoS One. 2011;6:e27421. doi:10.1371/ per night is unchangedÞ journal.pone.0027421. 9. Fischer S, Drosopoulos S, Tsen J, Born J. Implicit learning – explicit knowing: a role for sleep in memory system interaction. J Cogn Neurosci. 2006;18:311–319. Data availability doi:10.1162/089892906775990598. 10. Karni A, Tanne D, Rubenstein BS, Askenasy JJ, Sagi D. The CHARLS data are publicly available at http://charls.pku. Dependence on REM sleep of overnight improvement edu.cn/en. of a perceptual skill. Science. 1994;265:679–682. doi:10. 1126/science.8036518. 11. United States Census Bureau. An aging world: 2015. Disclosure statement https://www.census.gov/content/dam/Census/library/ No potential conflict of interest was reported by the author. publications/2016/demo/p95-16-1.pdf.PublishedMarch, 2016. Accessed June 3, 2019. 12. Ellenbogen JM, Payne JD, Stickgold R. The role of Ethical approval sleep in declarative : passive, permissive, active or none? Curr Opin Neurobiol. The Lewis University Institutional Review Board (IRB) 2006;16:716–772. doi:10.1016/j.conb.2006.10.006. approved this study. 13. Potkin KT, Bunney WE Jr., García AV. Sleep improves memory: the effect of sleep on long term memory in early adolescence. PLoS One. 2012;7:e42191. doi:10. Funding 1371/journal.pone.0042191. 14. Golinowska S, Groot W, Baji P, Pavlova M. Health This research received no specific grant from any funding promotion targeting older people. BMC Health Serv agency in the public, commercial, or not-for-profit sectors. Res. 2016;16:345. doi:10.1186/s12913-016-1514-3. 15. Sindi S, Kåreholt I, Johansson L, et al. Sleep distur- bances and dementia risk: a multicenter study. ORCID Alzheimers Dement. 2018;14:1235–1242. doi:10.1016/j. Wei Chen http://orcid.org/0000-0001-5255-9293 jalz.2018.05.012. 16. Shokri-Kojori E, Wang GJ, Wiers CE, et al. β-Amyloid accumulation in the human brain after one night of References sleep deprivation. Proc Natl Acad Sci. 2018;115:44 83–4488. doi:10.1073/pnas.1721694115. 1. Knutson KL, Ryden AM, Mander VA, Van Cauter E. 17. Li X, Jiang Q, Li S, Feldman MW. Female fertility Role of sleep duration and quality in the risk and history and mid-late-life health: findings from China. severity of type 2 diabetes mellitus. Arch Intern Med. J Women Aging. 2018;30:62–74. doi:10.1080/08952841. 2006;166:1768–1774. doi:10.1001/archinte.166.16.1768. 2016.1259445. 2. Taheri S. The link between short sleep duration and 18. Tham S, Chen Y, Li L, Zhang J, McLaughlin S. obesity: we should recommend more sleep to prevent Successful aging, adaptation and resilience. Innov obesity. Arch Dis Child. 2006;91:881–884. doi:10.1136/ Aging. 2018;2:163–164. doi:10.1093/geroni/igy023.590. adc.2005.093013. 19. Li Q, Wang Y, Zhao Y. The impact of China’s new 3. Heinzelmann M, Lee H, Rak H, et al. Sleep restoration rural pension program on elderly labor, grandchild is associated with reduced plasma C-reactive protein care, and old-age support. Fem Econ. 2018; and depression symptoms in military personnel with 24:265–287. doi:10.1080/13545701.2017.1421768. sleep disturbance after deployment. Sleep Med. 20. Lin B, Lin Z, Zhang YY, et al. The impact of the new 2014;15:1565–1570. doi:10.1016/j.sleep.2014.08.004. rural pension scheme on retirement sustainability in 4. Parikh PK, Troyer AK, Maione AM, Murphy KJ. The China: evidence of regional differences in formal and impact of memory change on daily life in normal aging informal labor supply. Sustainability. 2018;10:1–7. and mild cognitive impairment. Gerontologist. doi:10.3390/su10124366. 2015;56:877–885. doi:10.1093/geront/gnv030. 21. Cheng HG, Chen S, McBride O, Phillips MR. 5. Maquet P. The role of sleep in learning and memory. Prospective relationship of depressive symptoms, Science. 2001;294:1048–1052. doi:10.1126/science.1062 drinking, and tobacco smoking among middle-aged 856. and elderly community-dwelling adults: results from 6. Stickgold R. Sleep-dependent memory consolidation. the China Health and Retirement Longitudinal Study Nature. 2005;437:1272–1278. doi:10.1038/nature04286. (CHARLS). J Affect Disord. 2016;195:136–143. doi:10. 7. Squire LR, Zola SM. Episodic memory, semantic mem- 1016/j.jad.2016.02.023. ory, and . . 1998;8:205–211. 22. Kesavayuth D, Liang Y, Zikos V. An active lifestyle and doi:10.1002/(SICI)1098-1063(1998)8:3<205::AID- cognitive function: evidence from China. J Econ Ageing. HIPO3>3.0.CO;2-I. 2018;12:183–191. doi:10.1016/j.jeoa.2018.05.001. SLEEP TIME AND MEMORY 373

23.SunY,ShiL,BaoY,SunY,ShiJ,LuL.The 30. Redline S, Kirchner HL, Quan SF, Gottlieb DJ, bidirectional relationship between sleep duration Kapur V, Newman A. The effects of age, sex, ethnicity, and depression in community-dwelling middle-aged and sleep-disordered breathing on sleep architecture. and elderly individuals: evidence from a longitudinal Arch Intern Med. 2004;164:406–418. doi:10.1001/ study. Sleep Med. 2018;52:221–229. doi:10.1016/j. archinte.164.4.406. sleep.2018.03.011. 31. Horne J. Sleepfaring: The Secrets and Science of a Good 24. Wang TY, Wu Y, Wang T, Li Y, Zhang D. Night ’sSleep. New York, NY: Oxford University Press; 2007. A prospective study on the association of sleep dura- 32. Tonetti L, Fabbri M, Natale V. Sex difference in sleep- tion with grip strength among middle-aged and older time preference and sleep need: a cross-sectional sur- Chinese. Exp Gerontol. 2018;103:88–93. doi:10.1016/j. vey among Italian pre-adolescents, adolescents, and exger.2018.01.009. adults. Chronobiol Int. 2008;25:745–759. doi:10.1080/0 25. Yan M, Fu Z, Qin T, et al. Associations of sleep dura- 7420520802394191. tion and prediabetes prevalence in a middle-aged and 33. Dijk DJ. Regulation and functional correlates of slow elderly Chinese population with regard to age and wave sleep. J Clin Sleep Med. 2009;5:S6–S15. hypertension: the China health and retirement long- 34. Santhi N, Lazar AS, McCabe PJ, Lo JC, Groeger JA, itudinal study baseline survey. J Diabetes. Dijk D-J. Sex differences in the circadian regulation of 2018;10:847–856. doi:10.1111/1753-0407.12662. sleep and waking cognition in humans. Proc Natl Acad 26. Burn KF, Henderson VW, Ames D, Dennerstein L, Sci. 2016;113:E2730–E2739. doi:10.1073/pnas.1521637113. Szoeke C. Role of grandparenting in postmenopausal 35. American Academy of Sleep Medicine. Weekend women’s cognitive health: results from the women’s sleep fails to improve performance, but women han- healthy aging project. Menopause. 2014;21:1069–1074. dle workweek sleep loss better. https://aasm.org/week doi:10.1097/GME.0000000000000236. end-sleep-fails-to-improve-performance-but-women- 27. Janke K, Propper C, Shields MA. Assaults, murders handle-workweek-sleep-loss-better/.PublishedJune and walkers: the impact of violent crime on physical 9, 2011. Accessed July 29, 2019. activity. J Health Econ. 2016;47:34–49. doi:10.1016/j. 36. Goldstein JM, Jerram M, Poldrack R, et al. Sex differ- jhealeco.2016.01.006. ences in prefrontal cortical brain activity during fMRI 28. Tanaka S. Environmental regulations on air pollution in of auditory verbal . Neuropsychology. China and their impact on infant mortality. JHealthEcon. 2005;19:509–519. doi:10.1037/0894-4105.19.4.509. 2016;42:90–103. doi:10.1016/j.jhealeco.2015.02.004. 37.RångtellFH,KaramcheduS,AnderssonP,etal. 29. Mallampalli MP, Carter C. Exploring sex and gender A single night of sleep loss impairs objective but not differences in sleep health: a society for women’s health subjective working memory performance in a research report. J Women’s Health. 2014;23:553–562. sex-dependent manner. J Sleep Res. 2019;28:e12651. doi:10.1089/jwh.2014.4816. doi:10.1111/jsr.12651.